WP2: Continuous adjoint solvers

Objectives

• Improvement of numerical stability of the OpenFOAM adjoint discretisation, improvement of numerical accuracy.
• Advances with adjoint turbulent modelling on coarser meshes as typical in industrial application
• Provide a unified extensible code base for WP4-6

Methodology

Numerical stability issues will be addressed in a more targeted way without seriously impacting accuracy by NTUA, ENGYS and VW. The approach centres around identifying the sources of instability as they relate to specific characteristics of the computational grid and addressing the worst problems by either modifying the mesh generation process or damping/deactivating source terms in marginal elements. To enhance stability with respect to minor transients in the primal, a solution based on an ensemble of partially converged adjoint solutions will be explored by ENGYS. NTUA will incorporate the new techniques in adjoint-based truncated Newton methods as well.

The advances in WP2 support the application WPs 4-6.

Lead: NTUA, participants: ENGS, VW [months 1-36]

Participating Research Fellows

• Giorgos Karpouzas (ESR 4 at ENGYS): Topology Optimisation based in OPENFOAM Continuous Adjoint Solver (WPs 1, 2, 5, 6)
• Mehdi Ghavami Nejad (ESR 8 at NTUA): OpenFOAM-based continuous adjoint method for aerodynamic optimization of unsteady turbulent flows. (WPs 2, 3, 5)
• Christos Vezyris (ESR 8 at NTUA): Discrete and continuous adjoint methods: GPU-enabled and one-shot adjoints, for unsteady flow optimisation (WPs 2, 3, 4, 5)
• Nikolaos Magoulas (ESR 13 at VW): Application of continuous and discrete adjoint OpenFOAM to unsteady car aerodynamics, Adjoint-based optimisation workflows (WPs 1, 2, 3, 5)